关键词: 电力负荷变幅, 气象因子, GAM, 逐步回归, 多元线性回归

Abstract: Based on daily maximum power load in Shijiazhuang from 2017 to 2019 and meteorological data during the same period, the variation characteristic of daily maximum power load was analyzed. The correlation between variation of daily maximum power load and meteorological factors, air quality index (AQI) was also analyzed. The forecasting models of daily maximum power load variation in winter and summer were established by using stepwise regression, multiple linear regression and generalized additive model (GAM), the data of corresponding time in 2019 were taken as the independent test samples of forecasting effect. The results show that the daily maximum power load in Shijiazhuang had an obvious increasing trend from 2017 to 2019, and the correlation between variation of power load and factors had obvious seasonality. The factors which had a negative correlation with daily maximum power load variation in winter were positively correlated with daily maximum power load variation in summer, and vice versa. Among the three models, GAM model had the best prediction effect, and its forecast effect in summer was better than that in winter. In business application, GAM model with AQI could be selected in summer, and GAM model without AQI should be selected in winter.

Key words: power load variation, meteorological factor, GAM, stepwise regression, multiple linear regression